Binding affinity of transforming growth factor-beta for its type II receptor is determined by the C-terminal region of the molecule.

Transforming growth factor-β (TGF-β) isoforms have differential binding affinities for the TGF-β type II receptor (TβRII). In most cells, TGF-β1 and TGF-β3 bind to TβRII with much higher affinity than TGF-β2. Here, we report an analysis of the effect of TGF-β structure on its binding to TβRII by using TGF-β mutants with domain deletions, amino acid replacements, and isoform chimeras. Examination of the binding of TGF-β mutants to the recombinant extracellular domain of TβRII by a solid-phase TGF-β/TβRII assay demonstrated that only those TGF-β mutants containing the C terminus of TGF-β1 (TGF-β1-(Δ69-73), TGF-β1-(Trp71), and TGF-β2/β1-(83-112)) bind with high affinity to TβRII, similar to native TGF-β1. Moreover, replacement of only 6 amino acids in the C terminus of TGF-β1 with the corresponding sequence of TGF-β2 (TGF-β1/β2-(91-96)) completely eliminated the high affinity binding of TGF-β1. Proliferation of fetal bovine heart endothelial (FBHE) cells was inhibited to a similar degree by all of the TGF-β mutants. However, recombinant soluble TβRII blocked the inhibition of FBHE cell proliferation induced by TGF-β mutants retaining the C terminus of TGF-β1, consistent with the high binding affinity between these TGF-β molecules and TβRII. It was further confirmed that the TGF-β2 mutant with its C terminus replaced by that of TGF-β1 (TGF-β2/β1-(83-112)) competed as effectively as TGF-β1 with 125I-TGF-β1 for binding to membrane TβRI and TβRII on FBHE cells. These observations clearly indicate that the domain in TGF-β1 responsible for its high affinity binding to TβRII, both the soluble and membrane-bound forms, is located at C terminus of the molecule.

Transforming growth factor-␤s (TGF-␤s) 1 are multifunctional peptides that regulate many cellular processes, including growth, differentiation, inflammation, immunosuppression, and expression of extracellular matrix proteins (1)(2)(3)(4). The biologically active mammalian TGF-␤s are dimeric molecules (usually homodimers) consisting of disulfide-linked monomers that are derived by cleavage of the 112 C-terminal amino acids from a precursor polypeptide (1,3). Three distinct isoforms of TGF-␤ (␤1, ␤2, and ␤3) have been identified in mammalian cells (1). They show remarkable structural homology between each other, including 9 conserved cysteine residues that form four intrachain disulfide bonds (Cys 7 -Cys 16 , Cys 15 -Cys 78 , Cys 44 -Cys 109 , and Cys 48 -Cys 111 ) and one interchain disulfide bond at Cys 77 , and high amino acid sequence identity: Ͼ98% among the same isoforms across species and 71-76% among different isoforms. The existence of these highly conserved TGF-␤ isoforms across species implies very important specific roles for each isoform in vivo. Indeed, although TGF-␤ isoforms have a similar receptor-binding pattern and are indistinguishable in most in vitro biological assays, recent studies indicate that the TGF-␤ isoforms can also have selective actions in appropriate systems. For example, TGF-␤1 and TGF-␤3, but not TGF-␤2, strongly inhibit the growth of some endothelial cells (5,6) and hematopoietic cells (7,8). Similarly, TGF-␤1 inhibits growth of the human colorectal cancer cell line LS513, while TGF-␤2 has almost no effect on the growth of these cells (9). In contrast, TGF-␤2 and TGF-␤3, but not TGF-␤1, inhibit the survival of cultured chick ciliary ganglionic neurons (10). The differences in potency of TGF-␤ isoforms may correlate with specificity in the interactions between these isoforms and their cell-surface receptors.
The signaling receptors for TGF-␤, type I receptors (T␤RI) (53 kDa) and type II receptors (T␤RII) (70 -85 kDa), both contain extracellular ligand-binding, transmembrane, and intracellular Ser/Thr kinase domains (11,12). It has been shown that T␤RI and T␤RII associate as interdependent components of a heteromeric complex: T␤RI requires T␤RII to bind ligand, and both receptors are required for inhibition of cell growth and gene expression by TGF-␤ (13)(14)(15)(16)(17). TGF-␤ binding to T␤RII is required to initiate the TGF-␤ signaling pathway. Transfection of T␤RII into the human breast cancer cell line MCF-7 lacking TGF-␤ receptors results in the appearance of not only T␤RII, but also T␤RI, suggesting that T␤RI requires T␤RII for transport to the cell surface or TGF-␤ binding, or both (18).
Two other cell-surface TGF-␤-binding proteins of known structure are betaglycan (also called type III receptor (T␤RIII)) (250 -350 kDa) and endoglin (190 kDa) (19 -23). Betaglycan is a membrane-bound proteoglycan having a large extracellular domain and a very short cytoplasmic tail (41 amino acids) with no apparent signaling motif. Betaglycan is involved in presentation of TGF-␤s to T␤RI and T␤RII for signaling (24). Endoglin is a disulfide-linked homodimeric glycoprotein composed of two subunits of 95 kDa, showing ϳ70% structural homology to betaglycan at the transmembrane and short cytoplasmic regions of each subunit (22). It is expressed at high levels in vascular endothelial cells (21). The function of endoglin is currently unclear, although its expression is suggested to be critical to placental development (25), and mutation of the endoglin gene has been linked to hereditary hemorrhagic telangiectasia type 1 (26).
Accumulated evidence has shown that TGF-␤ receptors and membrane-binding proteins have different binding affinities for the TGF-␤ isoforms. In general, T␤RI and T␤RII bind more efficiently to TGF-␤1 and TGF-␤3 than to TGF-␤2 (27,28). Betaglycan has approximately equal binding affinity for all three TGF-␤ isoforms (24), whereas endoglin binds only to TGF-␤1 and TGF-␤3, but not to TGF-␤2 (22). The difference in receptor binding affinity between TGF-␤ isoforms may result from conformational differences in their structure or from the specificity of amino acid side chains. TGF-␤ isoforms share 71-76% identity in their amino acid sequence. A comparison of the solution structure of TGF-␤1 with the crystal structure of TGF-␤2 has indicated that the backbone structures of TGF-␤1 and TGF-␤2 are almost identical. 2 There are, however, several regions of variation in conformation and/or mobility between these isoforms that may be related to their functions (29,30). In addition, distribution of charged residues on the protein surface varies considerably (30). In our previous studies, TGF-␤ mutants including isoform chimeras, amino acid substitutions, or deletions were used to identify the specific amino acids (positions 45 and 47) in TGF-␤2 responsible for its high affinity binding to ␣ 2 -macroglobulin (␣ 2 M) (31,32) and another region (amino acids 69 -73) in TGF-␤1 responsible for its binding to receptors on LS513 colon carcinoma cells (33). In this study, we have applied the same techniques and utilized recombinant soluble human T␤RII to delineate the specific region(s) in the TGF-␤ molecule responsible for its high affinity binding to T␤RII. Our results show that a C-terminal domain containing amino acids 92, 94, and 98 of TGF-␤1 is necessary for specifying its high affinity binding to T␤RII.

Cell Culture
Chinese hamster ovary cells, NIH 3T3 cells, Mv1Lu mink lung epithelial cells, and fetal bovine heart endothelial (FBHE) cells were purchased from American Type Culture Collection (Rockville, MD) and cultured as described (31,34). Mv1Lu and FBHE cell growth inhibition assays (34,35) were used to determine the bioactivity of TGF-␤ mutants and the ability of soluble T␤RII to block TGF-␤ activity.

Synthesis and Purification of Soluble Extracellular Domain of T␤RII
The cDNA for the extracellular domain (amino acids 1-159) of the human TGF-␤ type II receptor (36) was expressed in NSO mouse myeloma cells or a baculovirus expression system and purified by affinity chromatography using either a TGF-␤1 column or a monoclonal anti-human soluble T␤RII antibody column (37). The purity of the protein was verified by SDS-polyacrylamide gel electrophoresis and silver staining.

Solid-phase TGF-␤/T␤RII Binding Assays
Two solid-phase TGF-␤/T␤RII binding assays were used to measure the binding affinity of TGF-␤s for T␤RII.
TGF-␤ Antibody Detection Method-This method was established based on the sandwich enzyme-linked immunosorbent assay for TGF-␤. All of the reaction buffers have been described previously (38). 96-well plates (Nunc Maxisorb, Nunc Inc.) were coated with 100 l/well of 0, 50, 100, 250, 500, or 1000 ng/ml soluble T␤RII at 4°C overnight and then blocked with 250 l/well of 10 mg/ml bovine serum albumin in blocking buffer. TGF-␤s at concentrations ranging from 0 to 20,000 pg/ml were added to the coated plates and incubated at room temperature for 1 h. After extensive washing, 100 l of antibodies (100 ng/ml) to TGF-␤1 or TGF-␤2 conjugated with horseradish peroxidase (R&D Systems, Minneapolis, MN) was added to the plates and incubated at room temperature for another hour. The plates were washed with washing buffer and further with distilled water. Peroxidase substrate 3,3Ј,5,5Ј-tetramethylbenzidine solution (1-Step TM Turbo TMB, Pierce) was added to the plates (100 l/well) to react with horseradish peroxidase at room temperature for 30 min and further developed by adding 100 l of 1 N H 2 SO 4 to produce a bright yellow solution. TGF-␤ binding was quantified by measurement of the absorbance of the solution at 450 nm.
Binding Competition Assay Using Biotinylated TGF-␤1-Reaction buffers have been described (38). 96-well Nunc Maxisorb plates were coated with 100 l of 50 ng/ml soluble T␤RII at 4°C overnight and then blocked with bovine serum albumin as described above. TGF-␤s at serially diluted concentrations ranging from 0 to 500 ng/ml mixed with a fixed amount (5 ng/ml) of biotinylated TGF-␤1 (R&D Systems) were added to the T␤RII-coated plates and incubated at room temperature for 1 h. After extensive washing, 100 l/well avidin-conjugated alkaline phosphatase (Pierce) in 1:1000 dilution in Tris-buffered saline (0.3% crystalline bovine serum albumin, 10 mM Tris, 150 mM NaCl, pH 7.4) was added to the plates and incubated at room temperature for 1 h. The plates were washed with washing buffer followed by distilled water. 100 l/well alkaline phosphatase substrate, p-nitrophenyl phosphate (1 mg/ml in 10 mM diethanolamine solution) (Kirkegaard & Perry Laboratories, Inc., Gaithersburg, MD), was added to the plates and incubated at room temperature until a bright yellow color developed. The signal from biotinylated TGF-␤1 bound to T␤RII was determined by measurement of the absorbance of the reaction mixture at 405 nm.

FBHE Cell Growth Inhibition Assay
The FBHE cell growth inhibition assay (34) was performed in 0.2% fetal bovine serum medium, in which TGF-␤1 and TGF-␤2 showed similar activity on inhibition of FBHE cell growth (32). In this assay, 10 pM TGF-␤s, which inhibits 90% incorporation of [ 3 H]thymidine (DuPont NEN), was added to 0.25 ϫ 10 5 FBHE cells in 48-well dishes together with increasing concentrations of soluble T␤RII ranging from 24 to 6000 ng/ml.

Differential Binding Affinity of TGF-␤ Isoforms for Soluble
Type II Receptor-TGF-␤1 and TGF-␤3 bind with much higher affinity than TGF-␤2 to the type II receptor (28). To identify specific high affinity epitopes of TGF-␤1, we utilized the recombinant extracellular domain of the human type II receptor for solid-phase TGF-␤/T␤RII binding assays. We first measured the binding of native TGF-␤1 and TGF-␤2 in this assay. In each experiment, the plates were coated with an increasing amount (5, 10, 25, 50, or 100 ng/well) of T␤RII, and the binding of TGF-␤ to the receptor was quantitated using a specific antibody against TGF-␤1 or TGF-␤2. Each antibody detects the corresponding TGF-␤ isoform with the same efficiency, as demonstrated in enzyme-linked immunosorbent assay experiments. 2 As shown in Fig. 1, TGF-␤1 binds soluble T␤RII at all of the different coating concentrations, whereas under the same conditions, the binding of TGF-␤2 is detectable only at the highest coating concentration of T␤RII (100 ng/well). As a control, TGF-␤1 or TGF-␤2 was added to the plates without the coating of soluble T␤RII. No significant binding was detected, suggesting that TGF-␤ binding in this assay is T␤RII-dependent. These data are consistent both with the higher binding affinity of TGF-␤1 for T␤RII expressed on the cell membrane (28) and with the affinity of soluble T␤RII for TGF-␤ isoforms reported recently by Lin et al. (42), where the apparent dissociation constant (K D ) for binding to TGF-␤1 was ϳ200 pM, while that for TGF-␤2 was undetectable. Fig. 1 also shows that the binding of TGF-␤1 increases with the coating concentration of T␤RII in the range of 5-25 ng/well, but does not change significantly when the coating concentration increases further from 25 to 100 ng/well. Our interpretation of these results is that the amount of recombinant T␤RII coated on the plates at higher concentrations (25,50, and 100 ng/well) is in excess for TGF-␤1 binding and that, at these concentrations, the TGF-␤1-binding signal is determined by the TGF-␤1 concentration alone. In contrast, at lower coating concentrations (5 and 10 ng/well) of T␤RII, the TGF-␤1-binding signal is a function not only of the TGF-␤1 concentration, but also of the amount of recombinant T␤RII coated on the plates.
The binding of TGF-␤3 to recombinant T␤RII was not measured by this method because of lack of appropriate antibodies. However, we show below that TGF-␤3 binds to recombinant soluble T␤RII in a ligand binding competition assay with an affinity similar to that of TGF-␤1 (see Fig. 3D).

Blocking Effect of Soluble T␤RII on Inhibition of FBHE Cell Proliferation by TGF-␤ Mutants-
To evaluate further the importance of the C terminus of TGF-␤1 in binding to soluble T␤RII, we tested the ability of soluble T␤RII to reverse the inhibition of growth of FBHE cells by TGF-␤ chimeras modified at the C-terminal region. FBHE cells were selected for this assay because they express very little betaglycan (34), allowing soluble T␤RII to efficiently block access of TGF-␤1 and TGF-␤ mutants to cell membrane T␤RII and thus reverse the inhibition of FBHE cell proliferation induced by TGF-␤. In Mv1Lu cells, which express a high level of betaglycan, very high amounts of soluble T␤RII are required to block the activity of TGF-␤ (data not shown). In this study, 10 pM TGF-␤s, which results in ϳ90% inhibition of [ 3 H]thymidine incorporation in FBHE cells, was added to these cells together with increasing amounts of soluble T␤RII. Fig. 4 shows that recombinant soluble T␤RII efficiently blocks the inhibition of FBHE cell growth induced by TGF-␤1 and TGF-␤2/␤1-(83-112), but not by TGF-␤2 and TGF-␤1/␤2-(83-112). In the absence of added TGF-␤, the growth of FBHE cells is not affected by recombinant soluble T␤RII (data not shown). These data clearly indicate FIG. 1. Differential affinity binding of TGF-␤ isoforms to soluble T␤RII. 96-well plates were coated with 0, 5, 10, 25, 50, or 100 ng/well recombinant soluble T␤RII. Increasing concentrations of TGF-␤1 (A) or TGF-␤2 (B) were added to the plates for binding. TGF-␤ binding to T␤RII on the plates was detected using TGF-␤1or TGF-␤2-specific antibodies conjugated with horseradish peroxidase. Dose-response curves were measured by the absorbance of the horseradish peroxidase substrate 3,3Ј,5,5Ј-tetramethylbenzidine chromophore measured at 450 nm. All samples were tested in triplicate, and the standard error of TGF-␤ binding was Ͻ5% of the mean value for each sample.
that soluble T␤RII has high affinity only for those TGF-␤ molecules containing the C terminus (amino acids 83-112) of TGF-␤1.

Binding Competition of TGF-␤ Mutants with 125 I-TGF-␤1 for Membrane TGF-␤ Receptors on FBHE Cells-To confirm that the C terminus (amino acids 83-112) of TGF-␤1 is important
for TGF-␤1 high affinity binding to cell membrane T␤RII, receptor cross-linking competition experiments were performed. FBHE cells were cross-linked with 25 pM 125 I-labeled TGF-␤1 in the presence of 500, 250, 100, 50, or 25 pM nonradioactive TGF-␤1, TGF-␤2, or TGF-␤2/␤1-(83-112). Fig. 5 shows that TGF-␤1 and TGF-␤2/␤1-(83-112), but not TGF-␤2, compete with 125 I-labeled TGF-␤1 for binding to T␤RI and T␤RII on FBHE cells. In contrast, TGF-␤1/␤2-(83-112) is ineffective (data not shown). Since it is known that recruitment of T␤RI to the TGF-␤ receptor complex occurs only after binding of TGF-␤1 to T␤RII (13)(14)(15)(16)(17), competition of 125 I-labeled TGF-␤1 binding to T␤RI by nonradioactive TGF-␤s is dependent on the affinity of these TGF-␤s to bind T␤RII. These data thus confirm that the C terminus of TGF-␤1 is important for its high affinity binding not only to soluble T␤RII, but also to its membranebound form. DISCUSSION We have shown that the C-terminal region (amino acids 83-112) of TGF-␤1 is essential for the high affinity binding of TGF-␤1 to recombinant soluble T␤RII and have confirmed this finding in living cells. Since most cells express multiple TGF-␤ receptors, we have used recombinant soluble T␤RII, containing only the extracellular domain of this receptor, to investigate the binding pattern of the TGF-␤ mutants. This method is simple and straightforward and avoids interference from other TGF-␤-binding proteins such as betaglycan on the cell membrane.
Comparison of the crystal structure of TGF-␤2 (43,44) with the NMR solution structure of TGF-␤1 (29,30) has indicated that the structures of these two isoforms differ in several regions including amino acids 8 -12, 70 -76, and 90 -98. The C-terminal region (amino acids 83-112) of TGF-␤1 reported here, which specifies the high affinity binding of TGF-␤1 to T␤RII, includes one of these domains (amino acids 90 -98). Moreover, Flanders et al. (45) observed that among polyclonal antibodies raised against a series of synthetic peptides corresponding to different regions of TGF-␤1, only the antibody to peptide 78 -109 effectively blocked binding of TGF-␤1 to receptors. Recently, Postlethwaite and Seyer (46) reported that a  3. Competition of TGF-␤ mutants for binding of biotinylated TGF-␤1 to soluble T␤RII. T␤RII binding competition assays were performed using a fixed amount (5 ng/ml) of biotinylated TGF-␤1 and increasing amounts (0.5-500 ng/ml) of each TGF-␤ mutant on plates coated with T␤RII (5 ng/well). Biotinylated TGF-␤1 bound to T␤RII was measured by the absorbance at 405 nm, using alkaline phosphatase-conjugated avidin and p-nitrophenyl phosphate. The blocking effect of non-biotinylated TGF-␤ mutants on the binding of biotinylated TGF-␤1 was calculated as the percentage blocking relative to control wells without unlabeled TGF-␤ added. Each sample was tested in triplicate, and the standard error was Ͻ10% of the mean value. Competition curves of unlabeled TGF-␤1 and TGF-␤2 are presented in A-D for comparison of the ability of TGF-␤ mutants to compete for the binding of biotinylated TGF-␤1 to T␤RII. The competition curves of four groups of TGF-␤ mutants are shown. 7-residue peptide containing amino acids 89 -95 of TGF-␤1 or larger peptides containing these 7 residues, but not peptides representing other regions of TGF-␤1, stimulated chemotactic migration of neutrophils, monocytes, and foreskin fibroblasts. Collectively, these data strongly suggest that the C terminus of TGF-␤1 is an important domain for receptor binding.
The data here show that exchange of the C-terminal regions of TGF-␤1 and TGF-␤2 (i.e. TGF-␤1/␤2-(83-112) and TGF-␤2/ ␤1-(83-112)) can switch binding patterns of TGF-␤ isoforms for T␤RII, and TGF-␤1 and TGF-␤3 bind similarly to T␤RII. This suggests that amino acids identical in the C-terminal regions of TGF-␤1 and TGF-␤3, but different in TGF-␤2, might be respon-sible for the differential binding affinities of TGF-␤ isoforms for T␤RII. Comparison of the amino acid sequences of the C-terminal regions (amino acids 83-112) of the three TGF-␤ isoforms shows that there are only 3 amino acids (positions 92, 94, and 98) that fit into this category. In TGF-␤1 and TGF-␤3, residues 92, 94, and 98 are valine, arginine, and valine, respectively; in TGF-␤2, they are isoleucine, lysine, and isoleucine, respectively (Fig. 6B). Confirming this hypothesis, we were able to demonstrate that the ability of TGF-␤1 to bind to soluble T␤RII was markedly reduced by substitution of only 6 amino acids (positions 91-96) of TGF-␤2 into TGF-␤1 (TGF-␤1/ ␤2-(91-96)). Whether these residues interact with T␤RII directly or whether they can indirectly affect the conformation of the T␤RII-binding site on TGF-␤ is still unknown. Even though the three amino acid substitutions in TGF-␤ isoforms at positions 92, 94, and 98 are very conservative, change(s) in side chains of these amino acids could affect the conformation of the whole protein. However, this is unlikely since we know that the backbone structures of TGF-␤1 and TGF-␤2 are almost identical. Also, it is noteworthy that residues 91-98 compose the most extended and flexible region in TGF-␤ and undergo rapid local internal motion, which is not found in any other regions (30). Thus, it is more likely that this domain is directly rather than indirectly involved in TGF-␤ binding to T␤RII. Whether all 3 amino acids (residues 92, 94, and 98) are critical for the high affinity binding of TGF-␤1 and TGF-␤3 to T␤RII is under further investigation.
Recently, it has been demonstrated that whereas TGF-␤1 binds to T␤RII expressed alone on cells, TGF-␤2 binds only to cells on which both T␤RI and T␤RII are coexpressed, and then with an affinity comparable to that of TGF-␤1 binding (47). These data are consistent with the equal ability of TGF-␤1 and TGF-␤2 to inhibit proliferation of FBHE cells and the selective ability of soluble T␤RII to compete for TGF-␤s containing the C terminus of TGF-␤1. These data also suggest that the confor- ) and increasing amounts of soluble T␤RII (24 -6000 ng/ml) were used to perform the FBHE cell growth inhibition assay. The blocking effect of soluble T␤RII on FBHE cell growth inhibition induced by TGF-␤ was calculated as the percentage blocking relative to control wells with 10 pM TGF-␤, but no soluble T␤RII added. Each TGF-␤ sample was tested in triplicate, and the standard error was Ͻ5% of the mean value. The interchain disulfide bond between 2 cysteine residues at amino acid 77 is marked. The amino acids are numbered as "n" on one monomer subunit and as "nЈ" on the other subunit. B, sequence comparison of human TGF-␤1, TGF-␤2, and TGF-␤3. Identical amino acids in the three isoforms are shown as dashes. Amino acids 92, 94, and 98, which are critical for high affinity binding of TGF-␤1 and TGF-␤3 to T␤RII, are underlined.
mation of this C-terminal region in TGF-␤2 might be constrained or restricted by its binding to T␤RI in such a way as to increase the affinity of this region for binding to T␤RII.
We have previously reported two other domains (amino acids 45-47 and 69 -73) involved in isoform-specific activities of TGF-␤. The first domain specifies the site for the high affinity binding of TGF-␤2 to ␣ 2 M (32). Replacement of amino acids 45 and 47 with the corresponding amino acids in TGF-␤1 abolished the high affinity binding of TGF-␤2 to ␣ 2 M. Later, it was confirmed by Webb et al. (48) that these amino acids are critical in determining the affinity of TGF-␤2 for binding to native ␣ 2 M, but not to ␣ 2 M-methylamine, a chemically modified form that mimics activated ␣ 2 M induced by proteinases in vivo. In the TGF-␤ structure, amino acids 45-47 (in TGF-␤1, Leu 45 -Gly 46 -Pro 47 ; and in TGF-␤2, Ala 45 -Gly 46 -Ala 47 ) connect the Cterminal end of a ␤-strand (␤3, amino acids 38 -44) and the N-terminal end of an ␣-helix (␣3, amino acids 57-68) (Fig. 6A). Since the backbone conformation of this domain in TGF-␤1 and TGF-␤2 is very similar, one possible mechanism by which ␣ 2 M may discriminate between TGF-␤1 and TGF-␤2 is by recognizing this specific exposed hydrophobic surface area of TGF-␤2, but not that of TGF-␤1.
The other domain (amino acids 69 -73) is important for receptor binding in the colorectal cancer cell line LS513 (33). Deletion of amino acids 69 -73 of TGF-␤1 dramatically decreased its binding to TGF-␤ receptors in this cell line, but not on other cells such as Mv1Lu cells and another colorectal cancer cell line, LS1034. The TGF-␤/T␤RII binding data presented here clearly indicate that this domain is not responsible for the high affinity binding of TGF-␤1 to T␤RII. Therefore, it is more likely that this region may be critical to target a unique or possibly mutated receptor, not yet characterized on LS513 cells. Also, it is noteworthy that the growth of LS513 cells is inhibited by TGF-␤1, but is refractory to TGF-␤2 (9). Comparison of the solution structure of TGF-␤1 with the crystal structure of TGF-␤2 indicates that residues 69 -72 adopt a type II ␤-turn in TGF-␤1 and a type I ␤-turn in TGF-␤2 (29,30). The difference of only 1 amino acid in this ␤-turn (Gly-71 in TGF-␤1 and Glu-71 in TGF-␤2), which is located at the tip of an exposed loop in the TGF-␤ structure, could possibly affect the conformation of this domain.
It is not known whether both C termini of TGF-␤1 on two monomer subunits are required for high affinity T␤RII binding. Heterodimers such as TGF-␤1.2 (27,49) and TGF-␤2.3 (49), purified from natural sources, will be good candidates for this study. We have recently shown that recombinant [Ser 77 ]TGF-␤1 was capable of binding to soluble T␤RII, but that fully chemically reduced TGF-␤1 was unable to bind (50). Whether [Ser 77 ]TGF-␤1 binds to soluble T␤RII by formation of noncovalently bound dimers, which could be stabilized by the extensive hydrophobic interfaces of two monomers, as shown for a similarly mutated form of platelet-derived growth factor (51), remains to be proven.
The biological effect of a particular TGF-␤ isoform on a given target cell is the product of 1) the selective "presentation" of the isoform to the different binding receptors, governed by the accessory proteins including membrane proteins (e.g. betaglycan) and extracellular proteins (e.g. ␣ 2 M); and 2) the signaling pathway triggered by the receptor combination that is activated by the TGF-␤ isoform presented in this manner. Although many models have been proposed for the interactions between TGF-␤ receptors and between TGF-␤ ligands and receptors (52,53), they are all based on data from experiments in which receptors have been overexpressed on the surface of cells transfected with respective cDNAs. The stoichiometry of complexes of TGF-␤ receptors with or without ligand in vivo is still not well understood. The goals of our research are to identify the specific regions of TGF-␤ involved in binding to its signaling receptors as well as to accessory proteins that control the accessibility of TGF-␤ to its signaling receptors. The strategy we have reported here involves study of T␤RII as a single component for analysis of ligand binding in vitro and then extends the observations in more complex cellular systems. Applying this strategy to other TGF-␤ receptors and accessory proteins will elucidate the complex interactions among ligands, receptors, and other binding proteins, which underlie the different biological potencies of TGF-␤ isoforms. Moreover, the demonstration of similar three-dimensional structures for TGF-␤2 and osteogenic protein 1, a distantly related member of the TGF-␤ superfamily with ϳ30% amino acid identity to TGF-␤s, suggests that these structure-function studies of TGF-␤s will also identify important receptor-binding epitopes in other members of the TGF-␤ superfamily (54).